https://bugs.winehq.org/show_bug.cgi?id=38177
Bug ID: 38177 Summary: d3d: Card 10de:0849 not found in driver DB Product: Wine Version: 1.7.37 Hardware: x86 OS: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: directx-d3d Assignee: wine-bugs@winehq.org Reporter: bernhardu@vr-web.de Distribution: Debian
(Opening this ticket as I got suggested in bug #33969.)
When starting "Volvo - The Game" with wine 1.7.37 I get such a line: err:d3d:init_driver_info Card 10de:0849 not found in driver DB.
Later the game complained also about too less graphics memory.
Therefore I tried raising it by using the registry key "VideoMemorySize" to 256 and that made the game work.
Unfortunately because this is an onboard graphics card with shared memory I do not know exactly which value would be correct.
Due to dmesg the nouveau driver uses "VRAM: 256 MiB".
The board is this one: http://www.asrock.com/mb/NVIDIA/K10N78M/ There the graphics card is described as: "GeForce 8100" with "Max. shared memory 512MB".
Also it mentions the graphic device is a Geforce 8100 (10de:084f), but wine gets the pci id 10de:0849 which is currently not in use. (And 10de:0849 would be the Geforce 8200.)
This all is while using the nouveau driver from current Debian/testing. (Because the nvidia binary driver 340 from debian made the system unstable.)
Attached is the output of dmesg, lspci -nnn and a patch to have 10de:0849 recognized with 256MB.
https://bugs.winehq.org/show_bug.cgi?id=38177
--- Comment #1 from Bernhard Übelacker bernhardu@vr-web.de --- Created attachment 50931 --> https://bugs.winehq.org/attachment.cgi?id=50931 Output of dmesg command
https://bugs.winehq.org/show_bug.cgi?id=38177
--- Comment #2 from Bernhard Übelacker bernhardu@vr-web.de --- Created attachment 50932 --> https://bugs.winehq.org/attachment.cgi?id=50932 Output of lspci command
https://bugs.winehq.org/show_bug.cgi?id=38177
--- Comment #3 from Bernhard Übelacker bernhardu@vr-web.de --- Created attachment 50933 --> https://bugs.winehq.org/attachment.cgi?id=50933 patch to add GeForce 8200 with 256MB
https://bugs.winehq.org/show_bug.cgi?id=38177
Matteo Bruni matteo.mystral@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |matteo.mystral@gmail.com
--- Comment #4 from Matteo Bruni matteo.mystral@gmail.com --- (In reply to Bernhard Übelacker from comment #3)
Created attachment 50933 [details] patch to add GeForce 8200 with 256MB
Good catch. If you can temporarily install the binary Nvidia drivers and check what renderer string is reported by it and update the cards_nvidia_binary table (the "8200" entry should either use CARD_NVIDIA_GEFORCE_8200 or be removed altogether) that would be excellent.
Apparently the CARD_NVIDIA_GEFORCE_8500GT device ID also suffers from the same issue. You could add an entry for it too in gpu_description_table if you can find all the info to put there. Or just remove the define and replace its use with e.g. the 8600GT one. This could be a separate patch.
https://bugs.winehq.org/show_bug.cgi?id=38177
--- Comment #5 from Bernhard Übelacker bernhardu@vr-web.de --- Created attachment 50944 --> https://bugs.winehq.org/attachment.cgi?id=50944 Output of glxinfo while running the binary nvidia driver
https://bugs.winehq.org/show_bug.cgi?id=38177
--- Comment #6 from Bernhard Übelacker bernhardu@vr-web.de --- Created attachment 50945 --> https://bugs.winehq.org/attachment.cgi?id=50945 Output of glxinfo while running the free nouveau driver
https://bugs.winehq.org/show_bug.cgi?id=38177
--- Comment #7 from Bernhard Übelacker bernhardu@vr-web.de --- With binary nvidia driver installed: OpenGL renderer string: GeForce 8100 / nForce 720a/integrated/SSE2/3DNOW!
But what still makes me wonder: - lspci shows 10de:084f - wine gets somehow 10de:0849 (which would be a GeForce 8200)
https://bugs.winehq.org/show_bug.cgi?id=38177
--- Comment #8 from Matteo Bruni matteo.mystral@gmail.com --- (In reply to Bernhard Übelacker from comment #7)
With binary nvidia driver installed: OpenGL renderer string: GeForce 8100 / nForce 720a/integrated/SSE2/3DNOW!
But what still makes me wonder:
- lspci shows 10de:084f
- wine gets somehow 10de:0849 (which would be a GeForce 8200)
It doesn't matter. I assume you put CARD_NVIDIA_GEFORCE_8200 in the "8100" entry of the cards_nvidia_binary table. That's what makes wined3d recognize the 8100 as a 8200 and that's fine.
Thank you for taking the time to look this up too.
https://bugs.winehq.org/show_bug.cgi?id=38177
--- Comment #9 from Bernhard Übelacker bernhardu@vr-web.de --- That is already the case without my modification.
As far as I can see wine gets this PCI Id from wined3d_guess_card which gets only as input the vendor and renderer string. (There the function select_card_handler queries cards_nvidia_mesa.)
Therefore as long as nouveau returns in the renderer string "NVAA" wine treats this card as 8200 with 10de:0849.
Then the question remains how many memory should be given: - the spec for this board mentions 512MB (are there 8200/8100 with just 256?) - nouveau prints in dmesg it (initially?) took 256MB
https://bugs.winehq.org/show_bug.cgi?id=38177
--- Comment #10 from Matteo Bruni matteo.mystral@gmail.com --- (In reply to Bernhard Übelacker from comment #9)
That is already the case without my modification.
As far as I can see wine gets this PCI Id from wined3d_guess_card which gets only as input the vendor and renderer string. (There the function select_card_handler queries cards_nvidia_mesa.)
Therefore as long as nouveau returns in the renderer string "NVAA" wine treats this card as 8200 with 10de:0849.
Ah, sure, for Nouveau that's going to be the case, since we can't disambiguate between the 8100 and the 8200. That's fine.
Then the question remains how many memory should be given:
- the spec for this board mentions 512MB (are there 8200/8100 with just 256?)
- nouveau prints in dmesg it (initially?) took 256MB
It doesn't matter much, both are fine. I'd just go with 512MB, it's less likely to break games.
https://bugs.winehq.org/show_bug.cgi?id=38177
--- Comment #11 from Bernhard Übelacker bernhardu@vr-web.de --- Created attachment 50955 --> https://bugs.winehq.org/attachment.cgi?id=50955 tried to find out where the NVAA in the render string comes from, seems from the board itself.
https://bugs.winehq.org/show_bug.cgi?id=38177
--- Comment #12 from Bernhard Übelacker bernhardu@vr-web.de --- Patch sent: https://www.winehq.org/pipermail/wine-patches/2015-March/137599.html
https://bugs.winehq.org/show_bug.cgi?id=38177
--- Comment #13 from Bernhard Übelacker bernhardu@vr-web.de --- Patch got commited: http://source.winehq.org/git/wine.git/commit/44ddcac6c372b003c81e36aa9f55bf3...
https://bugs.winehq.org/show_bug.cgi?id=38177
Bruno Jesus 00cpxxx@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Fixed by SHA1| |44ddcac6c372b003c81e36aa9f5 | |5bf32f5c0f8bb Status|UNCONFIRMED |RESOLVED Resolution|--- |FIXED
--- Comment #14 from Bruno Jesus 00cpxxx@gmail.com --- (In reply to Bernhard Übelacker from comment #13)
Patch got commited: http://source.winehq.org/git/wine.git/commit/ 44ddcac6c372b003c81e36aa9f55bf32f5c0f8bb
Fixed.
https://bugs.winehq.org/show_bug.cgi?id=38177
Alexandre Julliard julliard@winehq.org changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |CLOSED
--- Comment #15 from Alexandre Julliard julliard@winehq.org --- Closing bugs fixed in 1.7.39.